1.数据导入
1.1导入facebook_location_train_set(数据集大小:1.8G), 需要加载一段时间
import pandas as pd
locdata=pd.read_csv(r"C:\Users\鹰\Desktop\ML_Set\FaceBook_train.csv\FaceBook_train.csv")
1.2对数据的信息的简单分析
显示部分数据
locdata.head()
查看数据描述
locdata.describe()
查看数据行列数
locdata.shape
2.数据基本处理
2.1缺失值处理
print(“查看数据缺失值:”)
print(locdata.isna().sum())
locdata.dropna()
print(locdata.isna().sum())
2.2数据提取
缩小数据范围
locdata= locdata.query(“x>2.0 & x<2.5 & y>2.0 & y<2.5”)
选择时间特征, 对time进行转化
time=pd.to_datetime(locdata[“time”], unit=“s”)
time=pd.DatetimeIndex(time)
locdata[“day”]=time.day
locdata[“hour”]=time.hour
locdata[“weekday”]=time.weekday
去掉签到较少的地方, 在这里去掉签到次数小于三的地点
place_set= locdata.groupby(“place_id”).count()
place_set= place_set[place_set[“row_id”]>3]
locdata=locdata[locdata[“place_id”].isin(place_set.index)]
确定目标值和特征值, 用loc和iloc可以吗? 有什么区别吗?
x_all=locdata[[“x”,“y”,“accuracy”,“day”,“hour”,“weekday”]]
y_all=locdata[“place_id”]####
2.3数据集分割
from sklearn.model_selection import train_test_split
x_train, x_test, y_train, y_test=train_test_split(x_all, y_all,test_size=0.2)###
print(x_train)
print(x_test)
print(y_train)
print(y_test)
3.特征工程
3.1特征预处理-标准化
from sklearn.preprocessing import StandardScaler
scaler=StandardScaler()
x_train=scaler.fit_transform(x_train)
x_test=scaler.fit_transform(x_test)
4.模型训练-KNN+CV
4.1KNN调用
from sklearn.neighbors import KNeighborsClassifier
estimator=KNeighborsClassifier()
4.2模型优化
from sklearn.model_selection import GridSearchCV
params={“n_neighbors”:[1,3,5,7,9]}
estimator=GridSearchCV(estimator, param_grid=params, cv=5)
4.3模型训练
estimator.fit(x_train, y_train)
5.模型评估
5.1预测值
y_predict=estimator.predict(x_test)
print(“预测值为:”, y_predict)
5.2准确率
score=estimator.score(x_test,y_test)
print(“准确率为:”, score)
5.2最优模型参数:
print(“最优模型为:”, estimator.best_estimator_)
5.3最好评分
print(“最高分:”, estimator.best_score_)
